Why does Perl warn about "useless constant 1" when using bigint?

There are two issues at hand here.

1. Why does use bigint; 1; warn in void context?
2. Why is the constant being executed in void context in the first place?

---

$ perl -c -we'1 while sub_with_side_effects();'
-e syntax OK

$ perl -c -we'use bigint; 1 while sub_with_side_effects();'
Useless use of a constant (1) in void context at -e line 1.
-e syntax OK

Why does use bigint; 1; warn in void context?

use bigint; installs a callback that's called when a constant literal is encountered by the parser, and the value returned by the callback is used as the constant instead. As such, under use bigint;, 1 is no longer truly just a simple 0 or 1.

But you're not doing anything wrong, so this warning is spurious. You can work around it by using () or undef instead of 1.

undef while sub_with_side_effects();

Unless I needed to use it throughout my code base, I would favour the following:

while ( sub_with_side_effects() ) { }

---

$ cat Module.pm
package Module;
use bigint;
1;

$ perl -c -w Module.pm
Useless use of a constant (1) in void context at Module.pm line 3.
Module.pm syntax OK

Why is the constant being executed in void context?

When Perl executes a module, Perl expects the module to return a scalar value, so Perl should be executing the module in scalar context.

However, you told Perl to compile the script Module.pm. When Perl executes a script, Perl doesn't require any values to be returned, so Perl executes the script in void context.

Using a module as a script can cause spurious warnings and errors, and so can passing -W. Test a module using as follows:

perl -we'use Module'

Actually, you shouldn't even need -w since you should already have use warnings; in the module. All you really need is

perl -e'use Module'

-W instead of use warnings; in your module or checking modules with -c instead of perl -MFoo -e0 can show spurious errors. This is an example of the latter.

When you load a module normally, it isn't in void context, because it is checking that the result is true.

(Note that when I try it using 5.20.1, the -W also results in a spurious overload arg '..' is invalid at /usr/share/perl/5.20/Math/BigInt.pm line 155.)